Machine Learning and Computational Modelling to define the Abnormal Heart Tissue responsible for Fatal Heart Rhythms
Key information
Start date: Between November 2026 and March 2027
Duration: Four years full-time. Stipend available for 3.5 years of the project.
Entry requirements: Applicants must hold a 2:1 undergraduate degree and/or a merit in a Master’s degree.
Eligibility: Open to UK students only
The project offers opportunities to work across multiple disciplines, including:
- Machine learning and artificial intelligence
- Biomedical signal processing
- Computational modelling
- Clinical electrophysiology
The student will join a highly collaborative environment involving clinicians, engineers, mathematicians, and computer scientists, with access to unique clinical datasets and state-of-the-art mapping technologies.
This PhD would particularly suit candidates with backgrounds in:
- Artificial intelligence or machine learning
- Computer science
- Data science or computational modelling
- Engineering
- Mathematics or applied mathematics
- Physics
About the project
Sudden cardiac arrest (SCA) accounts for approximately 15-20% of reported deaths in the UK. Mostly, these deaths are linked to cardiac arrhythmias. One of the most common and dangerous is Ventricular Tachycardia (VT), a rapid and unstable heart rhythm that often arises from regions of diseased or scarred heart tissue. VT can lead directly to sudden cardiac death if not treated immediately.
Project summary
VT can be treated with a procedure called ‘catheter ablation’, which significantly reduces VT recurrence rates, reduces hospitalisations, and improves survival compared with medication alone. The goal of VT ablation is conceptually simple: find where the fatal heart rhythm is originating and use either thermal or electrical energy to destroy, or ablate, that tissue, leaving the rest of the heart to function normally. However, procedural success rates of catheter ablation are modest. Improving the precision and effectiveness of VT ablation is therefore a major unmet clinical challenge.
This project is an interdisciplinary program that brings together Warwick Medical School and partners from the School of Engineering and Industry, combining machine learning and computational modelling to solve a cardiovascular medicine problem. The central aim is to develop next-generation computational approaches to identify the electrical signatures of diseased cardiac tissue and improve the targeting of VT ablation procedures.
During ablation, wires are introduced into the heart to examine the electrical properties of the tissue. These wires collect local electrical signals, known as an electrogram (EGM), at 1000s of locations in the heart. Currently, the EGMs are then described in relatively simple terms, for example, by their maximum amplitude or timing, and these features are then used to create a map of the heart to guide where to ablate. We call this electroanatomical mapping, and it is central to modern-day VT ablation. However, compressing these signals into 1 or 2 descriptive features means we are using only a fraction of the data we collect.
In this project, we will therefore seek to use advanced mathematical and machine learning approaches to identify hidden information within cardiac electrical signals that is currently unavailable to clinicians.
Methodology
To unravel the hidden information, the student will apply advanced computational approaches to large-scale clinical datasets collected during VT ablation procedures at University Hospitals Coventry and Warwickshire NHS Trust, an internationally recognised centre for ventricular arrhythmia management. The project will involve analysing thousands of EGMs recorded directly from human hearts and developing new methods to extract clinically meaningful information from them. Potential approaches include advanced signal decomposition and feature extraction, time–frequency and spectral analysis, entropy-based metrics, graph representations of cardiac conduction, and supervised, unsupervised, and deep learning approaches for classification of abnormal electrical activity within the heart.
Recent work from our group has already shown how novel signal processing and machine learning approaches can potentially improve VT ablation. For example, we have shown how changing signal features analysed dramatically alters performance of mapping for targeting the area of ablate (https://pubmed.ncbi.nlm.nih.gov/41569210/), and how by deriving over 40 features from the EGM a machine learning model can be created even from signals collected during a healthy heartbeat (https://pubmed.ncbi.nlm.nih.gov/40703139/), which is easily collected from all patients (unlike trying to map the VT itself which is dangerous and only possible in about a third of patients). This PhD project will build directly on this foundation to develop more sophisticated and automated approaches.
How to apply
You can submit an online application for MPhil/PhD in Medicine via this webpage. The deadline for applications is Wednesday 30 September 2026 and interview will be held around mid-October (date TBC). If you have any queries please contact wmsrdcoord@warwick.ac.uk.